As some may already know, I've gone into a lot of detail about Lockfits here:
http://www.markhennessy.co.uk/articl...rs.htm#lockfit
Apart from complete failures - which can happen to any transistor - and ignoring the beefier types used as output devices with no or minimal heat sinking, the usual failure mode for these is an increase in noise.
Whether this increase in noise is a problem depends on where in the circuit the transistor is. Obviously, the earlier in the signal path, the more noticeable it will be. It's common to find T1 is noisy, whereas it's much more unusual to find T2 being a problem because T2 would have to be a *lot* worse than T1 for it to be audible at the output. Simple, really.
Of course, Lockfits were used for many other things. The RP25B uses one as a voltage regulator. The RP75 uses a couple to drive the tuning meter. If any of these went noisy, would anyone know?
What happens to a noisy Lockfit? I don't know if it will carry on getting noisier; I don't know if it will eventually fail completely. If the root failure mode is moisture ingress or similar, then perhaps that will indeed happen. But so far, there is no real data out there. And regarding them as "change on sight" will ensure that remains the case. Against that, a balance has to be made - especially if the set is difficult to dismantle and work on - but in a Hacker, I'd definitely regard them as "innocent until proven guilty".
I understand that others will have different views.